The Vestbrona Carbonate Field (VCF) seep site is located on the mid-continental shelf of the Norwegian Sea, off the coast of Kristiansund, central Norway, at a water depth of about 270 m (63°28′N, 6°31′E; Figure 1A). In June 2020, we investigated the area with the remotely operated vehicle (ROV) Ægir 6,000 on board the R/V G.O. Sars (University of Bergen). Prior to seafloor inspection, free gas flares were mapped from the ship through acquisition of multibeam bathymetry and water column data using a Kongsberg EM302 system (Figure 1B). The site is characterized by massive carbonate precipitates in a generally hemipelagic environment with sediment largely composed of mud (Figure 2).

A vertical, downward facing high-definition (HD) video camera mounted to the bottom of the ROV was used to map two locations (Mosaic 1: 679 m² and Mosaic 2: 563 m²) in the VCF where active seepage was observed in the form of small gas flares (bubbling) and microbial mats (Figure 1). The ROV was maintained at an altitude of 2 m above the seafloor and moved slowly at a speed of about 0.5 knots in a lawn mower fashion, to ensure overlap between video lines. Still images were extracted from the video every 2 seconds with the free software FFmpeg (http://ffmpeg.org/) and time stamps were used to obtain the corresponding navigation data from the ROV. Resulting images and navigation files were loaded into Agisoft’s Metashape software (version 1.6.2 build 10247, 2020) to construct two georeferenced mosaics (Figure 3).

All visible features (i.e., visible to the naked eye), such as animals (hereafter referred to as ‘megafauna’), microbial mats, litter and substrates were manually marked in ArcGIS Pro 2.6.0 via the Editor tool in the two georeferenced mosaics (Figure 3). Individuals were marked as point feature classes, whereas colonial animals, encrusting sponges and substrates were outlined using the polygon feature class. Classification was based on visible morphology and made to the lowest possible taxonomic level. Both the size of megafauna and the resolution of seafloor images did not always allow for identification of taxa to species-level. Therefore, our faunal inventory consists of groups based on taxonomic ranks ranging from species to class. Biological features (e.g., sponges) that could not be assigned to any taxon due to high levels of morphological plasticity between and within species, were classified as “morphotypes”, with some morphotypes possibly including several species. Some organisms could not be assigned to any specific phylum and were thus classified as ‘Unknown’. A variety of fish were observed and marked in both mosaics. Most of them were clearly demersal and swimming rather slowly just above the seabed or appeared completely immobile which allowed for reliable quantification. However, Atlantic Cod (Gadus morhua) was seen swimming slightly higher in the water column and was highly motile, making it difficult to effectively enumerate. This species was thus only included in species counts, but not in the statistical analyses.

Similarly, the 5 different substrate types (carbonate crusts, soft sediment, dead Desmophyllum pertusum, coral rubble and dropstones; Figure 2) were at times difficult to differentiate either due to overlap (e.g., corals growing on hard substrates) or due to image quality. Identification was made as best as possible and for the most part, texture differences or angled corners were clearly visible enough to distinguish between the different substrates.

Densities for all taxa were calculated based on the spatial extents of the two mosaics. Colonies of individuals (i.e., Ascidiacea, small living D. pertusum colonies and Parazoanthus sp.) were considered as individuals to calculate densities. Average megafaunal density, taxa richness (S), Pielou’s evenness (J), and Shannon’s diversity (H) were calculated from density data for each substrate across both mosaics to obtain integral community characteristics. As the Shannon index (H) in itself does not give any information on true diversity of the community and as its entropy is highly nonlinear, the effective number of species (ENS) was additionally calculated from all Shannon indices (exp(H)) (Jost, 2006).

Density data was square root transformed to reduce the impact of highly abundant taxa (e.g., sponges) and the similarity between “hard” (i.e., carbonate crusts, dead D. pertusum, dropstones) and ‘unconsolidated’ (i.e., coral rubble, soft sediment) substrates was estimated using the Bray-Curtis index. Non-metric multidimensional scaling (nMDS) was used to reveal separate groups and results were verified through an analysis of similarity (ANOSIM). Finally, contributions of the different morphotaxa to the dissimilarity between substrates was investigated through a similarity percentage analysis (SIMPER). All data analyses and data visualizations were performed in R with the packages “vegan” version 2.5–7 (Oksanen et al., 2022), “ggplot2” version 3.3.5 (Wickham et al., 2023), “ggpubr” version 0.4.0 (Kassambara, 2023) and “ggsci” version 2.9 (Xiao et al., 2023).

Free gas was sampled with a gas sampler by the ROV near Mosaic 1. In addition, two push cores within Mosaic 1 (PC1-2) and two adjacent to Mosaic 2 (PC3-4) (see Figure 1 for locations), were taken and sampled immediately after recovery on deck. For methane (CH₄) headspace analyses in the pore water samples, holes with a diameter of 1.5 cm were drilled into the plastic liner at intervals up to 5 cm (2 cm–26 cm), starting at the sediment-water interface. It should be noted however, that only three samples could be taken for PC4 (at 2 cm and 5 cm depth) and one for PC1 (at 24 cm). Sediment plugs of 3 ml were taken using a 5 ml syringe with the luer tip removed. Each sediment sample was transferred to a 20 ml serum vial containing 2 glass beads and 6 ml NaOH (2.5%) to prevent microbial degradation. The vial was immediately closed with a septum and an aluminum crimp seal and stored at 4°C until further analyses.

Stable carbon isotopes of methane (C₁) and ethane (C₂), as well as hydrogen isotopes of methane were analysed at Hydroisotop GmbH, Germany. For the analyses an aliquot of the free and headspace gas was taken with a 10 ml gastight syringe and injected into 20 ml headspace vial filled with helium (He). In the purge&trap autosampler (MessTechnik GmbH) the content of the bottle is flushed with He and trapped 20 min on the absorption material at −120°C. After fast heating up to 200°C the gas mixture was transferred to the GC–MS – IRMS system (Thermo Fischer Scientific GmbH). The GC (Trace Ultra) separates C₁-C₄ gases from each other which were then transferred to the combustion/pyrolysis interface for conversion of hydrocarbons to CO₂ or H₂ for carbon and hydrogen stable isotope measurements using an isotope ratio mass-spectrometer (IRMS, DeltaV Advantage). The isotopic composition (δ¹³C and δ²H) is reported in ‰ (δ-values) against the international standards Vienna Pee Dee Belemnite (VPDB) for carbon and Vienna Standard Mean Ocean Water (VSMOW) for hydrogen, according to the following equations: δ 13 C V P D B =   13 C /   12 C s a m p l e −   13 C /   12 C V P D B   13 C /   12 C V P D B   x 1000   ‰ δ 2 H V S M O W =   2 H /   1 H s a m p l e −   2 H /   1 H V S M O W   2 H /   1 H V S M O W   x 1000   ‰

Porewater samples for sulfide (ΣHS = H₂S + HS⁻ + S²) concentrations were taken at a ∿2-cm-depth resolution (2 cm–39 cm) along the length of PC1 and PC4, starting at the sediment-water interface. Porewater was extracted in a temperature-controlled room (4°C) with acid-washed rhizon samplers and syringes, and 0.5–2 ml of porewater was preserved with Zn(OAc)2 onboard (<30 min after rhizons were disconnected) for further analyses in the lab. Samples were kept frozen all the time until analyses were conducted. The concentrations of total dissolved sulfide (ΣHS = H₂S + HS⁻ + S²) were determined by the iodometric method (US Environmental Protection Agency, method 9030 and Pawlak and Pawlak, 1999). Before the analyses, samples were centrifuged for 5–7 min at 2000 rpm to separate the ZnS precipitates from the residual porewater. The supernatant fluid was pipetted and discarded as it may contain other reductants (such as dissolved organic carbon species) that may react with I₂ and affect the results. The remaining ZnS precipitates were washed into a glass beaker with ca. 2 mL of 18Ω Milli-Q water for titration. Iodine (I₂) solution of ca. 14 mM was added (0.2 mL). Aliquots of starch solution (0.05 mL; prepared every other day) were added as an end-point indicator and 4M HCl was added to ensure a completed reduction of I₂ to 2I⁻ by lowering the sample pH with HCl (Pawlak and Pawlak, 1999). The ZnS in the sample then reduces I₂ when in contact (ZnS + I₂→ 2ZnI + S). We then titrated the residual I₂ to calculate the amounts of total sulfide in the samples (i.e., I_2-unreacted – I_2-residual = ZnS_sample). Factory-made 0.00109N Na-thiosulfate (stabilized standard solution, Hach Lot# 2408949) was sequentially diluted 10 times and 100 times and used as titrants. Titrants were added to the sample with an automatic pipette under constant mixing in an open beaker until the purple color faded away as a result of complete I₂ reduction. The amounts of titrant were then recorded for the calculation of ZnS_sample. As I₂ is fairly unstable when exposed to light, its concentration was closely monitored every ca. 30 min during the titration to constrain I₂-_unreacted. The uncertainty of the measurements was then determined from the two closest I₂ measurements before and after the titration of the actual sample. In general, the concentration of I₂-_unreacted decreased by 0.27 mM every hour. New I₂ was used during the same session of analyses if the I_2-unreacted concentration was below 85% of its concentration earlier in the session. The Zn-acetate solution used to precipitate out total sulfide was also titrated following the identical protocol to ensure no measurable sulfide remained in it.

All attached fauna from two carbonate rocks collected from within Mosaic 1 were retrieved and all organisms, e.g., including sponge fragments, brittle stars, polychaete fragments (Nephtys sp.) and cnidarians were immediately frozen at −20°C after recovery on deck. In the lab, the recovered organisms were freeze-dried in a vacuum chamber for 24 h, then dissected to separate soft and non-metabolically active tissues (e.g., muscle, tegument) or, when body size was small, were used whole (Mateo et al., 2008). They were subsequently ground to a homogeneous powder using mortar and pestle. Samples containing hard inorganic carbon parts that could not be physically removed were acidified by exposing them to HCl vapors for 48 h in an airtight container (Hedges and Stern, 1984). Stable isotope ratio measurements were performed via continuous flow - elemental analysis - isotope ratio mass spectrometry (CF-EA-IRMS) at University of Liège (Belgium), using a vario MICRO cube C-N-S elemental analyzer (Elementar Analysensysteme GMBH, Hanau, Germany) coupled to an IsoPrime100 isotope ratio mass spectrometer (Isoprime, Cheadle, United Kingdom). Isotopic ratios were expressed using the widespread δ notation (Coplen, 2011), in ‰ and relative to the international references Vienna Pee Dee Belemnite (for carbon), atmospheric air (for nitrogen) and Vienna Canyon Diablo Troilite (for sulfur). IAEA (International Atomic Energy Agency, Vienna, Austria) certified reference materials sucrose (IAEA-C-6; δ¹³C = −10.8 ± 0.5‰; mean ± SD), ammonium sulfate (IAEA-N-2; δ¹⁵N = 20.3 ± 0.2‰; mean ± SD) and silver sulfide (IAEA-S-1; δ³⁴S = −0.3‰) were used as primary analytical standards. Sulfanilic acid (Sigma-Aldrich; δ¹³C = −25.6 ± 0.4‰; δ¹⁵N = −0.13 ± 0.4‰; δ³⁴S = 5.9 ± 0.5‰; means ± SD) was used as secondary analytical standard. Standard deviations on multi-batch replicate measurements of secondary and internal lab standards (amphipod crustacean muscle) analyzed interspersed with samples (one replicate of each standard every 15 analyses) were 0.2‰ for both δ¹³C and δ¹⁵N and 0.3‰ for δ³⁴S.

Three video transects of the non-seep background area were extracted from the ROV video surveys of the surrounding background area. These videos were taken with a forward-facing and slightly downward tilted camera and since this camera setup does not allow for precise areal measurements to be made, georeferenced mosaics could not be constructed in the same manner as within the seep area. Nonetheless, time stamps were used to obtain the corresponding navigation data from the ROV which allowed for approximate areas of the seafloor covered by the transects to be calculated. All visible features (biological and non-biological, as in the two georeferenced mosaics) were counted within each transect and converted to densities based on the approximate area calculations. Diversity indices were then calculated for comparisons with the seep mosaics.

In total, 20,813 individuals/colonies of individuals were enumerated in Mosaic 1 (30.65 ind./m²) and 11,287 in Mosaic 2 (20.05 ind./m²) (Table 1; Figure 3). Forty-nine taxa (i.e., species, genus, family, class, and ‘morphotypes’, including microbial mats) across 9 phyla were identified within the two mosaics: Porifera, Bryozoa, Cnidaria, Annelida, Platyhelminthes, Mollusca, Echinodermata, Arthropoda and Chordata. These, however, are likely underestimates due to the difficulty of identifying fauna from images, the possible presence of cryptic species and some morphotypes possibly including several species. Therefore, the total taxonomic richness of the mosaics is likely higher than the 49 taxa shown in Table 1. The majority of all marked individuals (M1: 95%; M2: 97%) belonged to the phylum Porifera and included most importantly cushion-like orange sponges (possibly of the genus Suberites), globular/irregular white/cream colored sponges (possibly of the genera Geodia and Mycale), encrusting orange sponges (possibly of the genus Hymedesmia), encrusting yellow/orange sponges (possibly of the genus Amphilectus) and foliaceous white/cream colored sponges (possibly of the genera Phakellia and Axinella) (Figure 4). All remaining morphotaxa (i.e., not belonging to the phylum Porifera) together, only accounted for 5% or less of all marked individuals in both mosaics. Sponges were thus by far the most dominant members of the megafaunal community at the VCF seep. Furthermore, besides being extremely abundant, many sponges also displayed considerable sizes and very diverse three-dimensional morphologies (e.g., Phakellia sp., Axinella sp., Geodia sp., glass sponge, Figure 5). Following sponges, the next most numerous taxa were unidentifiable ‘fan’ animals (possibly bryozoans), Munida squat lobsters, Echiurids and Parazoanthus polyps (Figure 6). Among the various taxa identified in the mosaicked area, several also hold commercial value, such as Atlantic cod (G. morhua), common ling (Molva molva) and tusk (Brosme brosme) (Kvangarsnes et al., 2012; Helle et al., 2015; Norwegian Directorate of Fisheries, 2020). Despite this being a seep site, microbial mats appeared rather small (in the range of several cm at the most) and no visible chemosynthesis-based animals such as siboglinid polychaete worms were seen.

Five substrates were observed at the VCF: carbonate crusts, soft sediment, dead D. pertusum, coral rubble and dropstones. Of these, soft sediment made up the largest proportion of the surface area in both mosaics (M1: 69% and M2: 65%) (Table 1). Regarding hard substrata, M1 contained a large amount of carbonate crusts (27% of mosaicked area) but very little coral rubble (3% of mosaicked area), while M2 contained more coral rubble than carbonate crusts (22% and 12% of mosaicked area respectively). Dropstones and dead D. pertusum were less abundant and only accounted for less than 1% each of the total surface area in both mosaics (Table 1).

Based on the nMDS and ANOSIM, the fauna associated with the different substrates was significantly different (R: 0.67; p = 0.03). Faunal assemblages of ‘hard’ (i.e., carbonate crusts, dead D. pertusum, dropstones) were also significantly different from those on ‘unconsolidated’ (i.e., coral rubble, soft sediment) substrates (R: 0.91; p < 0.01) (Figure 7). In terms of megafaunal densities, carbonate crusts were the most highly populated substrate in both mosaics (M1: 108 ind./m²; M2: 135 ind./m²), dominated however, by extremely high numbers of sponges (e.g., the cushion-like orange sponge or the globular/irregular white/cream colored sponge) leading to low taxa evenness and thus only intermediate diversity values compared to the other substrates (Table 2). Nonetheless, carbonate crusts hosted the absolute majority of all marked individuals (M1: 95%; M2: 82%), as well as the majority (M1: 91%; M2:89%) of all morphotaxa identified in both mosaics. In contrast, soft sediment, which dominated both mosaics, hosted only 3%–4% of all marked individuals and 57%–60% of all morphotaxa identified. While the taxonomic composition on dead D. pertusum (M1: 58 ind./m²) and dropstones (M1: 55 ind./m²; M2: 80 ind./m²) was similar to that on carbonate crusts (66% and 60% similarity respectively) (Table 3), megafauna was only about half as dense and diversity values were comparatively low. Megafaunal densities were the lowest on coral rubble (M1: 2; M2: 13 ind./m²) and soft sediment (M1: 2; M2: 3 ind./m²), while diversity values were the highest among these two substrates. Community composition was very similar between soft sediment and coral rubble (57%), while similarities in taxonomic composition between soft sediment and most other substrates (i.e., carbonate crusts, dead D. pertusum and dropstones) were very low, ranging from 23% to 27%.

Stable carbon and hydrogen isotope analyses were conducted on the free gas samples to assess whether the underlying source of methane was of microbial (−110‰ ≥ δ¹³C ≥ −50‰ and −400‰ ≥ δD ≥ −150‰) or thermogenic (−50‰ ≥ δ¹³C ≥ −20‰ and −275‰ ≥ δD ≥ −100‰) origin (Whiticar, 1999). Wetness was measured by the molar ratio of methane to the sum of ethane and propane (C1/(C2+C3)), with wetter gases containing higher amounts of C2+ gases and displaying a low C1/(C2+C3) ratio. Microbial methane tends to be light and dry, whereas thermogenic methane is usually rather heavy and wet (Milkov and Etiope, 2018). The free gas was nearly entirely composed of methane (99.97%) with δ¹³C-CH₄ and δD-CH₄ values of −74.4‰ and −194‰ respectively, and with a ratio of methane over ethane and propane of >4,000, appears to be of microbial origin (Figure 8) (Whiticar, 1999; Nicot et al., 2017). Similarly, headspace gas analysis from the push cores also revealed methane of microbial origin (δ¹³C-CH₄: −79.1‰ – −91.7‰; δD-CH₄: −189‰ – −215‰ across all push cores). Note that while the δ¹³C-CH₄ values clearly confirm the hydrogenotrophic methanogenesis (CO2-reduction), also in the revised genetic diagram by Milkov and Etiope (2018), the isotopic composition of ethane (d13C-C2H6)of −37.2‰ is well within the thermogenic range (Nakagawa et al., 2003). We suggest two possible pathways to explain this: 1) microbial methanogenesis in shallow sediment as a result of organic matter degradation fueled by methane-seep related biomass or 2) secondary methanogenesis as a result of thermogenic hydrocarbon biodegradation (Stagars et al., 2017) or a combination of both processes. At least, some admixture of thermogenic gas in our free gas sample is evident from the high ethane δ¹³C value. In push cores PC2 and PC3, low concentrations of methane (3 and 10 µM respectively) were measured shallow in the sediment (2 cm), although concentrations increased downcore, but nonetheless remained below the millimolar range (Figure 9). On the other hand, in both push cores 1 and 4, millimolar concentrations (1.0 and 1.9 mM) of methane were measured at 24 and 2 cm depth respectively. Sulfide concentrations were in the low millimolar range (<0.6 mM) in the first sediment layers (0–3 cm), and rapidly increased downcore, with high millimolar concentrations (max: 1.1 and 1.2 mM respectively) measured across both PC1 and PC4 (Figure 9).

All δ³⁴S values ranged from 18.9 to 21.6‰ and were slightly lower or similar to ocean sulfate (MacAvoy et al., 2003) (Figure 10A). Bulk-tissue δ¹³C values varied between −18.3 and −25.8‰, corresponding to the range typically associated with biomass produced through photosynthesis (Hobson et al., 1996; Søreide et al., 2006), while δ¹⁵N values ranged between 7.7 and 8.2‰ for sponges, between 8.6 and 9.0‰ for cnidarians, between 9.3 and 9.7‰ for brittle stars and between 11.8 and 13.1‰ for polychaetes (Figure 10B) (All stable isotope data are published open access in SEANOE Sinner et al., 2020). These values are in line with samples belonging to different trophic levels, and food sources, e.g., filter feeding by sponges, deposit feeding by brittle stars and predation and deposit feeding by polychaetes (Fredriksen, 2003; Grall et al., 2006).

The three transects outside the seep area contained a total of 12 megafaunal taxa (Table 1) across 6 phyla (Porifera, Cnidaria, Mollusca, Echinodermata, Arthropoda and Chordata). Three phyla (Bryozoa, Annelida and Platyhelminthes), which were seen in the seep mosaics were absent from the background transects. On the other hand, two sponge morphotaxa (yellow globular and white/cream fistular sponges), as well as three fish species (Trisopterus esmarkii, Merluccius merluccius and one fish of unknown genus) were identified in the background area that were not seen in the seep areas. The holothuroid Stichopus tremulus accounted for the majority of all megafauna in the background transects, representing 44% of all marked individuals (T1 – T3 combined). Substrate features included large burrows or circular depressions in the soft sediment and small holes, the latter also being present in the seep mosaics. No hard substrata were recorded in the background transects. Mean megafaunal density as well as total taxa richness were notably lower in the background non-seep transects compared to the seep mosaics (25 ± 7.4 ind./m² vs. 0.15 ± 0.0 ind./m² and 48 taxa vs. 12 taxa) (Figure 11). However, while evenness was high in the background area (J: 0.86 ± 0.1) and low in the seep area (J: 0.47 ± 0.1), diversity indices were similar in both areas (H’: 1.73 ± 0.1 in the seep area vs. 1.49 ± 0.4 in the non-seep area), with the true diversity showing 6 ± 0.7 effective species in the seep area and 5 ± 1.6 effective species in the non-seep area.

Many instances of human-mediated litter were observed in the seep mosaics (Table 1; Figure 12). In total, 35 and 9 items were marked in M1 and M2 respectively, which consisted predominantly of fishing gear, but also glass, metals, paper, etc. (Figure 12). Fishing nets were the most abundant, with seven nets being marked in M1 and five in M2. Some reached sizes in the order of some square meters and many were entangled in soft corals and other benthic megafauna (Figure 12). Most of the litter did not appear to be particularly degraded, on the contrary, some bottles and cans were still perfectly intact. Considerably lower numbers of human litter items (2 in total) were observed in the transects of the background area outside the VCF seep.

Siboglinids dominate high latitude seeps and as chemosymbiotic fauna, they represent the base of the food chain and alter sediment geochemistry, functioning as ecosystem engineers (Levin, 2005; Cordes et al., 2010; Levin et al., 2016). Surprisingly, the VCF site did not contain any siboglinids at all. Low latitude shallow water cold seeps around the world tend to have a lower proportion of obligate taxa and a higher abundance of predators and other background taxa, than their deep water counterparts (Carney, 1994; Sahling et al., 2003; Tarasov et al., 2005; Vanreusel et al., 2009; Dando, 2010; Kiel, 2010). A possible hypothesis for this depth-related trend is that the increased input of photosynthetic organic matter from surface primary production at shallow depths mitigates the need for chemosynthesis as an energy source, even selecting against it and the energetically expensive adaptations it requires (Levin, 2005; Tarasov et al., 2005). Furthermore, shallow seeps are more prone to be invaded by higher levels of carnivorous predators from adjacent habitats, which would be able to efficiently prey on the largely sessile chemosynthesis-based (Carney, 1994; Sahling et al., 2003; Levin, 2005; Tarasov et al., 2005; Dando, 2010). Therefore, the relatively shallow water depth of the VCF site (270 m) could offer an explanation for the lack of siboglinids there. However, paradoxically, siboglinids have been recorded at Arctic seeps at considerably lower water depths (<80 m) (Savvichev et al., 2018; Vedenin et al., 2020). It has been hypothesized that among shallow water locations, the presence or absence of chemosymbiotic animals is linked to surface primary production, such that highly productive regions (with presumably considerable deposition of phytodetrital material to the seafloor) leads to the exclusion of chemosymbiotic animals, whereas more oligotrophic regions with lower levels of phytodetrital deposition favor them (Åström et al., 2022). Based on this, an explanation for the absence of siboglinids from the VCF site could be its location within the highly productive Norwegian Sea.

However, it can also be argued that the VCF site is simply beyond the photic zone where chemosynthesis-based symbioses would be selected against. In this case, other reasons need to be considered for the lack of chemosymbiotic animals at VCF. Since siboglinids are obligately symbiotic with chemosynthetic bacteria, the availability of chemosynthetic energy sources such as methane and sulfide represents a major limiting factor. We mapped two distinct gas flares in the multibeam survey (Figure 1B) and collected bubbling gas via the ROV close to the presence of carbonate crusts. This indicates that seepage is actively occurring at the VCF site. Indeed, we measured methane concentrations reaching up to 1.9 mM as shallow as 2 cm below the sediment-water interface and sulfide concentrations reaching up to 1.2 mM in only 5 cm of depth in the sediments. Therefore, it seems unlikely that sulfide (and methane) concentrations are insufficient to support siboglinids at the VCF site. However, despite sulfide and methane concentrations reaching millimolar concentrations at shallow depths in the sediment, we did observe considerable variation across different push cores. Thus, fluid release and sediment concentrations could be highly heterogeneous over very short lateral distances. Microbial mats were very patchy and small (Figure 13), which further supports the idea of the fluid regime at VCF being characterized by point sources with little lateral diffusion. Sediment heterogeneity (e.g., grain size) can lead to small-scale flow channelization, creating preferential rapid flow channels for methane, decreasing its residence time in the sediment and thus bypassing AOM by methane-consuming microbial communities (Torres et al., 2002; Luff et al., 2004; Mahadevan et al., 2012; Wankel et al., 2012). We did, in fact, observe patches of coarse sediment in the mosaics (Figure 14). The highly localized point sources of methane emission, which can act as a “bypass shunt” of AOM (Wankel et al., 2012) and by extension sulfide generation, may thus be a possible explanation for the absence of siboglinid polychaetes at the VCF seep.

Cold seep sabellids have recently been found to host methane oxidizing bacteria on their crowns and a nutritional symbiotic association has been suggested (Goffredi et al., 2020). We observed sabellids and sabellid-like animals at the VCF seep, including, in some instances, directly above microbial mats (Figure 15). Therefore, it is conceivable that methane-based chemosynthetic associations occur at the VCF seep, even if sulfide-based siboglinid chemosynthetic symbioses are absent. Nonetheless, the possible instances of methanotrophy-based sabellids at VCF are scarce overall in the community, in contrast with the extremely high densities and abundances exhibited by siboglinids when they are present at high latitude seeps (Åström et al., 2016; 2019; Sen et al., 2018b; 2018a; 2019b; Vedenin et al., 2020).

A lack, or paucity of chemosymbiotic animals would subsequently suggest that chemosynthetically fixed carbon does not constitute a major part of the food web at the VCF seep. We opportunistically sampled megabenthic fauna from two carbonate rocks retrieved from the VCF seep and conducted stable isotope analyses on them in order to assess the role of, and to track chemosynthetically fixed carbon through the food chain. Photosynthesis and chemosynthesis favor the two stable isotopes of carbon (¹²C and ¹³C) differently, therefore the ratio of the two constitutes a record of the carbon source at the base of the food web (DeNiro and Epstein, 1978; Fry, 2006). Specifically, biomass produced through chemosynthesis is depleted in the heavy isotopes of carbon (¹³C), and thus result in lower δ¹³C values, in comparison to biomass derived from photosynthesis (Kennicutt et al., 1992; MacAvoy et al., 2003). Methane seep organisms relying on chemosynthesis-derived organic matter have δ¹³C values typically lower than −30‰ when this matter is synthesized through sulfide oxidation, and even lower (sometimes as low as −60‰) when synthesized through methane oxidation (Demopoulos et al., 2010; Yamanaka et al., 2015). All of the sampled taxa of the VCF seeps displayed δ¹³C values between −25.8 and −19.6‰, suggesting a predominantly photosynthetic source of carbon (Hobson et al., 1996; Søreide et al., 2006) (Figure 10). Similarly, δ³⁴S values were typical for ocean sulfate (19–21‰), further suggesting a lack of sulfide based chemosynthetic pathways (MacAvoy et al., 2003; Grey and Deines, 2005; Yamanaka et al., 2015) (Figure 10A). δ¹⁵N values display a notable tissue-diet shift, increasing by 2–4‰ with each trophic level (Minagawa and Wada, 1984; McCutchan Jr et al., 2003). The lowest values we found were 7.69 and 8.18‰ but given that those individuals also had the highest δ¹³C values, it is unlikely that they represent low level consumers feeding on chemosynthetic microbes. Chemosynthetic signatures have even been observed in higher trophic level organisms at Arctic seeps. For example, δ¹³C values as low as −31‰ have been recorded in predatory Nepthys worms at the Bjørnøyrenna crater field (Åström et al., 2019). We also sampled Nepthys worms, and they displayed the highest δ¹⁵N values (11.8–13.1‰, Figure 10B), which indicates that they are predatory at VCF as well. However, δ¹³C values among these worms were between −21.7 and −19.6‰ implying a photosynthetic origin of the ingested carbon.

Therefore, the combined results of our stable carbon, sulfur and nitrogen stable isotope analyses suggest that chemosynthesis does not play a major role in the benthic food web at the VCF seep, and that taxa primarily obtain their nutrition from phytodetrital material. It should be kept in mind however, that only a small fraction of the local fauna was subjected to stable isotope analyses, and that taxa with the highest potential for being chemosymbiotrophic, such as the previously mentioned sabellids were not analyzed. Additionally, we observed high numbers of echiurans (M1: 123 individuals; M2: 44 individuals), probably Bonellia viridis, a species that has previously been suggested to feed on microbial mats at the periphery of the Milos vent fields (Dando et al., 1995). Therefore, despite a lack or paucity of chemosymbiotic fauna, it is possible that chemosynthesis nonetheless plays a role in the food web and nutrition of the benthic community at the VCF site, and we simply could not detect it based on our sampling efforts. Furthermore, no sampling was done on sediment infauna and we thus cannot exclude the possibility of chemosymbiotrophic infauna (e.g., thyasirid bivalves), commonly seen at Arctic and subarctic seeps (Åström et al., 2017; Sen et al., 2018a), from being present at the VCF seep. Nonetheless, a lack of extensive microbial mat cover and the absence of siboglinid worms in combination with our isotope results together strongly suggest a minor role, if any, of chemosynthesis and chemosynthetically fixed carbon in the local food web and nutrition of the resident fauna of the VCF seeps. This is despite high sediment sulfide concentrations and free gas methane emissions. Therefore, the VCF represents an active seep within what is generally considered the deep sea with a predominantly background community. This implies that direct evidence of active seepage does not necessarily result in chemosynthesis-based seep communities in high latitude locations. If fluid flow pathways are highly localized and restricted to point sources, background communities can develop, supported through conventional phytodetrital nutritional pathways, even at water depths below the photic zone.

Low latitude cold seeps around the world have been seen to host high biomass but low diversity communities compared to the surrounding non-seep seafloor, while seeps in higher latitudes appear to host both high abundance and high diversity communities in comparison to surrounding non-seep areas (Fisher et al., 2007; Levin et al., 2016; Sen et al., 2018a; 2019a; Åström et al., 2020). Our results seem to fall in between both trends with the VCF seeps hosting high biomass communities, compared to the background area (25 ind./m² versus 0.15 ind./m²) with however probably similar diversity (Figure 11). Similar to other methane seep studies (Levin et al., 2016; Sen et al., 2018a; 2019a; Åström et al., 2018; 2020), we link the presence of most benthic community members and their high abundances to heightened habitat heterogeneity and the occurrence of hard substrates, especially carbonate crusts, which are absent from the surrounding non-seep background area (Figure 11). Carbonate crusts are formed due to an increased carbonate alkalinity in the seep sediment, mediated through microbial activity, specifically, through the anaerobic oxidation of methane (AOM) coupled to the reduction of sulfate, a highly localized process which can continue at seep locations over hundreds to thousands of years (Boetius et al., 2000; Joye et al., 2004; Bayon et al., 2009; Sen et al., 2019b).

Despite carbonate crusts not representing the most abundant substrate within the seep area, they hosted nearly all morphotaxa identified, as well as the majority of all marked individuals. In contrast, soft sediment, which dominated both mosaics, only hosted about half of the morphotaxa identified and less than a tenth of all marked individuals. Thus, even compared to other hard substrates, such as dead D. pertusum and dropstones, the latter also being found in the non-seep background area (in some supplementary video material, but not in the video transects), carbonate crusts seem to play a particularly important role in structuring the local benthos. This may be explained by methane-derived authigenic carbonate crusts having higher three-dimensional complexity, offering more nooks and crannies for different kinds of animals to find fixation points (e.g., Parazoanthus sp.) and shelter in (e.g., Munida sp.); higher elevation from the seafloor, giving heightened access to organic material from the water column (e.g., Sabellidae); and sheer volume/size compared to the other hard substrates such as dropstones or dead D. pertusum corals, offering more space for animals to grow in large numbers (e.g., cushion-like orange sponge) and to grow in size (e.g., Paragorgia arborea) (Gutt and Schickan, 1998; Meyer et al., 2016; Sedano et al., 2020). Moreover, while dropstones constitute island-like habitats for hardbottom-dwelling fauna at high latitudes, increasing taxa richness of the region overall, they are limited by their small size and consequently mechanisms such as hydrodynamics, affecting food supply, larval dispersal and recruitment (Meyer et al., 2016; Ziegler et al., 2017). Seep carbonate crusts on the other hand are much bigger and may thus function more like archipelagos (Sen et al., 2019a), where the previously mentioned limitations may not apply in the same manner. Instead, faunal recruitment on these substrates may be linked to the presence or absence of interacting species or their channelization of seeping fluids that could enhance or aggregate particular species (Grupe, 2014).

Though carbonates had the highest abundances and numbers of taxa overall, other substrates, and most importantly the two unconsolidated substrates (i.e., soft sediment and coral rubble) showed higher taxa evenness and diversity. In addition, community composition on these substrates highly differed from all other substrates through the absence or low abundance of hard-substrate dwelling suspension-feeding animals, such as sponges, sabellids, ‘fan animals’ or soft corals. Only one single taxon was unique to these substrates, S. tremulus, a soft-bottom dwelling sea cucumber commonly found on soft sediments along the Norwegian coast (Schagerström and Sundell, 2021). Differences between the two unconsolidated substrates were also observed, primarily the presence of a few more hard-substrate-dwelling animals, such as sponges and ‘fan animals’ and mobile animals such as Munida sp. and ophiuroids on coral rubble compared to soft sediment. Indeed, D. pertusum rubble has been linked to an increased abundance of predators, in contrast to soft sediment and increased diversity compared to hard substrates covered in microbial mats, sponges, or corals (Jonsson et al., 2004; Lessard-Pilon et al., 2010). Moreover, the diversity of microhabitats found on coral rubble has been found to provide perfect conditions for the settlement of juveniles and to protect them until adulthood (Jensen and Frederiksen, 1992). It is thus the multitude of different substrates, that makes the VCF different from the background seafloor, by offering a variety of colonization surfaces for the background taxa, supporting them from larval settlement to their adult life, and thereby functioning as an important megafaunal oasis and ‘ecological stepping stone’ that would not be possible if only soft sediment and dropstones would be present (Jensen and Frederiksen, 1992; Gutt and Schickan, 1998; Kiel, 2016; Meyer et al., 2016; Sedano et al., 2020).

Norway is the seventh biggest oil producer, the third largest net oil exporter, as well as the world’s second largest exporter of fish and seafood in the world (Mohn and Osmundsen, 2008; Lipková and Hovorková, 2018; Johansen et al., 2019). In order to limit the significant detrimental impacts these activities have on benthic communities, Norway has created its ‘Integrated Ocean Management Plans’, restricting the use of bottom trawls in areas with coral reefs (i.e., D. pertusum) and at depths exceeding 1,000 m and managing seismic surveying and exploration drilling in oil-bearing formations through seasonal permits (ICES, 2019; Norvegian Ministry of Climate and Environment, 2020). However, no targeted measures exist to protect sponge aggregations in water depths less than 1,000 m and/or coral garden habitats from human stressors unless they occur in conjunction with D. pertusum reefs in the closed areas (OSPAR, 2010). As a consequence, hard-bottom coral gardens, dominated by vulnerable and very fragile soft corals are still under high fishing pressure and their total area has considerably declined in the last decades (Norvegian Ministry of Climate and Environment, 2020).

The VCF seep, with its multitude of hard substrates, hosts a high diversity and high abundance faunal community that is distinct from the surrounding seafloor community. Based on OSPAR (The Convention for the Protection of the Marine Environment of the North-East Atlantic) descriptions, the VCF seep ought to be categorized as a threatened and/or declining habitat, as they not only feature D. pertusum colonies and soft corals, such as P. arborea, Paramuricea placomus and Primnoa resedaeformis, but also deep-sea sponge aggregations, including some which have reached very large sizes. All these taxa are extremely slow-growing and are therefore particularly vulnerable to disturbances, such as oil drilling, bottom trawling and other fisheries related activities (Clark et al., 2016; Sundahl et al., 2020). Deep-water coral and sponge habitats often have high levels of endemism, host the early life-stages of many deep-sea animals including juvenile fish of commercial value, act as substrate and shelter for many other species, and play a major role in marine biogeochemical cycles, acting as hotspots of carbon processing in the food devoid deep ocean (Jensen and Frederiksen, 1992; Buhl-Mortensen and Mortensen, 2004; Buhl-Mortensen and Mortensen, 2005; Rogers, 2004; Hogg, M.M. et al., 2010; Beazley et al., 2013; Kenchington et al., 2013; Cathalot et al., 2015; Maldonado et al., 2017; De Clippele et al., 2019; D’Onghia, 2019). Furthermore, sponges in conjunction with corals seem to be particularly important, as they have a huge pallet of functional roles within coral ecosystems (Wulff, 2001; McLean and Yoshioka, 2007; Bell, 2008; De Goeij et al., 2013). Destructive relationships range from boring sponges infesting the tissue-barren portions of coral skeletons, to sponges outcompeting corals for space and overgrowing them (Beuck et al., 2007; McLean and Yoshioka, 2007; Bell, 2008). Many sponge species, however, have a beneficial, rather than destructive, relationship with corals. Sponges are known to increase coral survival by binding live corals to the substrate and preventing access to their skeletons, to mediate the regeneration of physically damaged corals and to return dissolved organic carbon (DOC) from the water column to the coral reef ecosystem as detritus, while producing detrital particulate organic Nitrate (PON), which may act as high-quality food source for corals and other detritivores and suspension feeders (Wulff, 2001; McLean and Yoshioka, 2007; Bell, 2008; De Goeij et al., 2013; Pawlik and McMurray, 2020). In addition to functioning as a biogenic habitat and controlling coral populations, sponges are also highly effective filter-feeders and play an important role in denitrification, carbon sequestration and benthic–pelagic coupling (Pham et al., 2019; Rooks et al., 2020). The presence and activity of sponges and corals therefore likely enhances settlement of other animals at the VCF site, making it a high abundance and biomass location, despite a lack of typical seep ecosystem engineering taxa. Furthermore, cold water corals on the Norwegian shelf have been hypothesized to be linked to seeps (Hovland and Risk, 2003), and the VCF site represents a location where cold water corals are directly associated with seeps.

In the VCF seep mosaics we observed a considerable amount of marine litter (on average 0.03 items/m²), the majority of which was fishing gear (Table 1). Seven fishing nets were seen in M1 and five in M2, with one of the bigger fishing nets in M1 being completely entangled in a large P. arborea soft coral (Figure 12A). Demersal fishing such as bottom gillnetting or bottom trawling (De Juan and Lleonart, 2010) has thus taken place at the VCF at some point. Importantly as well, we observed much more fishing gear in the mosaics in comparison to the background areas (Table 1), even with differences in areas being accounted for. This could be due to the abundance of large carbonate slabs at the VCF seep, as well as associated large, three-dimensional structure-forming taxa, which stand in the way of nets that are towed along the ocean floor, leading to fragments of the nets being ripped off. Therefore, the seep itself could be more susceptible to being affected by fishing from a purely physical perspective.

Recovery time after damaging impacts (e.g., physical damage through fishing nets) in coral habitats may take decades and even though recovery time appears to be shorter for deep-sea sponge aggregations, they have been found to suffer from lingering damage and experience delayed mortality over the course of many years or even decades (Rogers, 2004; Althaus et al., 2009; Rooper et al., 2011; Clark et al., 2016; Malecha and Heifetz, 2017). Our results indicate that, in clear contrast with the surrounding non-seep background area, the VCF seeps with their multitude of substrates act as biomass hotspots, hosting a wide array of animals, including not only species of commercial interest but also large sponge aggregations and a variety of soft and hard corals. The VCF thus not only increases the diversity of the region overall, but also acts as a habitat for various species considered priorities for protection under the OSPAR convention, while possibly playing an important role in nitrogen cycling, carbon retention and benthic–pelagic coupling. Therefore, persistent fishing pressures, or the commencement of oil drilling could put animals that have lived there for decades, maybe centuries at risk, which would additionally affect associated benthic and epibenthic fauna. Our results therefore highlight why seeps should be maintained as particularly valuable and vulnerable habitats and protection efforts need to be implemented in order to achieve adequate protection and prevent significant adverse impacts, compromising ecosystem integrity.